Tuesday, December 02, 2003

Wasted Maths

NZpols wastes a lot of maths defending presumptive rule consequentialism. I say wasted, because she missed my point completely.

The retreat to rule consequentialism is driven primarily by clashes between consequentialism and justice. The classic example here is that of McCloskey, who points out that consequentialists (well, he talks about utilitarians - hedonic consequentialists - but the argument is applicable to general consequentialism unless an infinite disvalue is placed upon certain actions) must bear false witness against an innocent person if it that would stop a riot. Or use collective punishment to prevent guerilla activity in wartime. Or impose harsh punishments on the guilty to meet public demand for retribution. Or support laws designed to defend an insecure democracy by making it a crime to "arouse the suspicions of the government".

In other words, any arbitrary injustice is allowed provided it makes people overall better-off by whatever metric is used. This is widely regarded as a knock-down argument for act-consequentialism.

So, consequentialists retreat to rules. Instead of asking "would this act make people better off", they ask "what general rules of conduct will make people better off". A society where people routinely bear false witness against the innocent would be fairly unpleasant, so it's better not to. This solves the problem of justice, but at the cost of consistency - because, as NZPols points out, there will be cases where the rule clearly results in "bad decisions" (meaning people will be worse off by whatever metric is used).

The move to presumptive rule consequentialism - having rules, but ignoring them in cases where following them would make people worse off - restores consistency, but brings the problem of justice right back with it. After all, McCloskey's cases are precisely those where violating general rules (specifically those against against injustice) will result in people being better off. In fact, it's difficult to see how presumptive rule consequentialism differs from act consequentialism in any significant way; the only difference it makes is that you should be really, really, really sure about things before you execute an innocent man to please the mob.